Abstract
District heating and district cooling networks are sustainable technology to cover the heating and cooling needs of buildings in urban areas. A new generation of this technology, called "fifth generation", is a cutting-edge solution that is emerging in Europe. This technology operates at a temperature close to the ground and includes electric-driven heat pumps in the substation to satisfy the user comfort. This solution implies new strategies for controlling substations and decentralized thermal energy storage together with the possibility of exploiting a large amount of low-temperature waste heat sources available in urban areas. This research aims at presenting the state of the art of fifth-generation district heating systems and the implementation of advanced control strategies to exploit the thermal capacity of centralized and decentralized thermal energy storage systems. The first part of this thesis presents a literature review of 40 fifth-generation district heating networks that are in operation in Europe. The results show that Switzerland and Germany are the pioneer countries in this technology. The interest in these networks is documented by the fact that in the last decade in Europe at least three plants of this type have entered into operation per year.
The study analyses the crucial role of thermal energy storage in district heating networks both in the case of high-temperature networks to cover the peak load and in fifth-generation networks to implement "demand response" strategies by exploiting the heat pumps of the substations. These strategies could be implemented for example in order to reduce the user's bill or to provide support for the electricity grid in the event of a high generation of electricity from non-dispatchable renewable sources. To make the substations intelligent, it is important to analyse the development of advanced controllers. This thesis analyses the application of a predictive controller based on artificial intelligence algorithms. In fact, to make predictions the controller uses a reduced-order model of a substation. The model was developed in the form of artificial neural networks starting from a numerical model developed in TRNSYS and validated with dedicated laboratory tests at the Institute for renewable energies of EURAC Research. The predictive controller has been implemented in LabVIEW and has been tested its capacity in the prediction of the behaviour and performance of a substation and to manage it by trying to satisfy both the comfort limits for the user and the technological limits of the heat pump.